Matrixnet features&cuts.ipynb

{
 "metadata": {
  "name": "",
  "signature": "sha256:ad92f61202bd92753d27230887a47d1f4ba3c72e27a3586d0b8faa239302672e"
 },
 "nbformat": 3,
 "nbformat_minor": 0,
 "worksheets": [
  {
   "cells": [
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "import sys\n",
      "sys.path.insert(0, '/home/mikari/REP')\n",
      "sys.path.insert(0, '/local/userdata/shelf/notebook/axelr/lhcb_trigger_ml/IPythonWorkflow/')\n",
      "sys.path.insert(0, '/home/mikari/topo_code')\n",
      "import topo_utils\n",
      "import cern_utils\n",
      "from cern_utils import converter_csv\n",
      "\n",
      "train = converter_csv.load_from_csv('/local/userdata/shelf/notebook/mikari/topo/learn_data_updated_data/train', sep='\\t')\n",
      "test = converter_csv.load_from_csv('/local/userdata/shelf/notebook/mikari/topo/learn_data_updated_data/test', sep='\\t')\n",
      "\n",
      "train,test = topo_utils.append_labels([train, test])\n",
      "train = train.select('is_passed_l0_goodGenB==1')\n",
      "test = test.select('is_passed_l0_goodGenB==1')\n",
      "train = topo_utils.add_folding_field(train, 1000)\n",
      "test = topo_utils.add_folding_field(test, 1000)\n",
      "train_subset = cern_utils.data_storage.DataStorage.union([train.select('signal==1').select('fold < 1000'), train.select('signal==0')])\n",
      "# test_subset = cern_utils.data_storage.DataStorage.union([test.select('signal==1').select('fold < 300'), test.select('signal==0')])\n"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "empty_events = {\n",
      "    11114001: 164  ,\n",
      "    11124001: 49  ,\n",
      "    12103035: 421  ,\n",
      "    12165106: 273  ,\n",
      "    12265042: 53  ,\n",
      "    11874042: 129  ,\n",
      "    11296013: 23 ,\n",
      "    30000000: 107352  ,\n",
      "}\n",
      "\n",
      "test_empty_events = {}\n",
      "for key, value in empty_events.iteritems():\n",
      "    if (key == 30000000):\n",
      "        test_empty_events[key] = value * 0.5\n",
      "    else:\n",
      "        test_empty_events[key] = value * 0.5\n",
      "    \n",
      "print test_empty_events"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "variables = [\"sumpt\",\n",
      "             \"m\",\n",
      "             \"mcor\",\n",
      "             \"ipchi2\",\n",
      "             \"vchi2\",\n",
      "             \"sumipchi2\",\n",
      "             \"fdr\",\n",
      "             \"nlt16\",\n",
      "             \"minpt\",\n",
      "             \"ptau\",\n",
      "             \"q\",\n",
      "             \"eta\",\n",
      "             \"nhlt1\",\n",
      "             \"nmu\",\n",
      "             \"n\",\n",
      "             \"fdchi2\"]\n",
      "\n",
      "variables_small = [\"sumpt\",\n",
      "             \"m\",\n",
      "             \"mcor\",\n",
      "             \"ipchi2\",\n",
      "             \"vchi2\",\n",
      "             \"fdchi2\",\n",
      "             \"minpt\"]"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "reload(topo_utils)\n",
      "metric = topo_utils.metric_generate(test.get_data(), 2500.0 / 1e6, test_empty_events, metric_plot=True)\n",
      "metric_5000 = topo_utils.metric_generate(test.get_data(), 5000.0 / 1e6, test_empty_events, metric_plot=True)\n",
      "\n",
      "def minpt_cut(data):\n",
      "    return numpy.array(data['minpt'] > 500)\n",
      "\n",
      "metric_minptcut = topo_utils.metric_generate(\n",
      "    test.get_data(),\n",
      "    2500.0 / 1e6,\n",
      "    test_empty_events,\n",
      "    metric_plot=True,\n",
      "    additional_cut=minpt_cut)\n",
      "\n",
      "metric_minptcut_5000 = topo_utils.metric_generate(\n",
      "    test.get_data(),\n",
      "    5000.0 / 1e6,\n",
      "    test_empty_events,\n",
      "    metric_plot=True,\n",
      "    additional_cut=minpt_cut)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "a = minpt_cut(test.get_data())"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "a"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "train_sgn = train_subset.select('truematch==1').select(\"presel==1\").select(\"signal==1\")\n",
      "train_bkg = train_subset.select(\"presel==1\").select(\"signal==0\")"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "from cern_utils import ef_classifier"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "mn_full = ef_classifier.ClassifierEF(directory='mikari-topo/', formula_name='mikari_topo_all_features-new-4-fixed', \n",
      "                                     dataset_name='mikari_topo_all_features-new-4-fixed', url_base='w73.cern.yandex.net')\n",
      "mn_full.set_params(features=variables, iterations=5000, features_sample_rate_per_iteration=0.8, training_fraction=0.5, regularization=0.05, sync=False)\n",
      "mn_full.fit(train_sgn, train_bkg)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "mn_small = ef_classifier.ClassifierEF(directory='mikari-topo/', formula_name='mikari_topo_small_features-new-4-fixed', \n",
      "                                     dataset_name='mikari_topo_small_features-new-4-fixed', url_base='w73.cern.yandex.net')\n",
      "mn_small.set_params(features=variables_small, iterations=5000, features_sample_rate_per_iteration=0.8, training_fraction=0.5, regularization=0.05, sync=False)\n",
      "mn_small.fit(train_sgn, train_bkg)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "mn_minpt = ef_classifier.ClassifierEF(directory='mikari-topo/', formula_name='mikari_topo_minpt-new-4-fixed', \n",
      "                                     dataset_name='mikari_topo_minpt-new-4-fixed', url_base='w73.cern.yandex.net')\n",
      "mn_minpt.set_params(features=variables, iterations=5000, features_sample_rate_per_iteration=0.8, training_fraction=0.5, regularization=0.05, sync=False)\n",
      "mn_minpt.fit(train_sgn.select('minpt > 500'), train_bkg.select('minpt > 500'))"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "no_nmu_variables = list(variables)\n",
      "no_nmu_variables.remove(\"nmu\")"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "mn_no_nmu = ef_classifier.ClassifierEF(directory='mikari-topo/', formula_name='mikari_topo_no_nmu-new-4-fixed', \n",
      "                                     dataset_name='mikari_topo_no_nmu-new-4-fixed', url_base='w73.cern.yandex.net')\n",
      "mn_no_nmu.set_params(features=no_nmu_variables, iterations=5000, features_sample_rate_per_iteration=0.8, training_fraction=0.5, regularization=0.05, sync=False)\n",
      "mn_no_nmu.fit(train_sgn, train_bkg)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "test_subset_labels = test.get_data(['signal'])['signal']"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "mn_minpt_preds = mn_minpt.predict(test)\n",
      "mn_full_preds = mn_full.predict(test)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "mn_small_preds = mn_small.predict(test)\n",
      "mn_no_nmu_preds = mn_no_nmu.predict(test)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "eff = dict()\n",
      "eff['trained with minptcut'] = metric(test_subset_labels, mn_minpt_preds)\n",
      "eff['trained with minptcut (with cut on test)'] = metric_minptcut(test_subset_labels, mn_minpt_preds)\n",
      "eff['all features'] = metric(test_subset_labels, mn_full_preds)\n",
      "eff['all features (with cut on test)'] = metric_minptcut(test_subset_labels, mn_full_preds)\n",
      "eff['7 features'] = metric(test_subset_labels, mn_small_preds)\n",
      "eff['7 features (with cut on test)'] = metric_minptcut(test_subset_labels, mn_small_preds)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "eff['no nmu'] = metric(test_subset_labels, mn_no_nmu_preds)\n",
      "eff['no nmu (with cut on test)'] = metric_minptcut(test_subset_labels, mn_no_nmu_preds)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def plot_eff(data_dict, p_label='p_empty', total_label='total_empty', alpha=1, move=3, bar_width=1):\n",
      "    plt.subplots(figsize=(13,7))\n",
      "    color = ['b', 'g', 'cyan', 'pink', 'orange', 'black', 'yellow', 'brown']\n",
      "    num = 0\n",
      "    bars = len(data_dict) + 1\n",
      "    for key, value in sorted(data_dict.iteritems()):\n",
      "        move = num\n",
      "        error_config = {'ecolor': '0.3'}\n",
      "        modes = sort(list(set(value.keys())))# - {30000000}))\n",
      "        channel_count = len(modes)\n",
      "        index = np.arange(channel_count)\n",
      "        plt.bar(bars * index + move, [1 - value[mode][p_label] for mode in modes], bar_width,\n",
      "                       alpha=alpha, color=color[num], \n",
      "                       yerr=[sqrt(value[mode][p_label] * (1 - value[mode][p_label]) / value[mode][total_label])\n",
      "                                                   for mode in modes], error_kw=error_config,\n",
      "                       label=key)\n",
      "        \n",
      "        num += 1\n",
      "        plt.xticks(bars * index + move * bar_width, modes)\n",
      "    plt.xlabel('Modes')\n",
      "    plt.ylabel('Inefficiency')\n",
      "    plt.title('Classifier Inefficiency for Matrixnet')\n",
      "    plt.legend(loc='best')\n",
      "    plt.tight_layout()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "2500HZ, With and without minpt cut"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "plot_eff(eff)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAA58AAAHyCAYAAABs5HrRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAABJRU5ErkJggg==\n",
       "text": [
        "<matplotlib.figure.Figure at 0x33ef7210>"
       ]
      }
     ],
     "prompt_number": 45
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "cmp_eff = dict()\n",
      "cmp_eff['all features 2500'] = metric(test_subset_labels, mn_full_preds)\n",
      "cmp_eff['all features 5000'] = metric_5000(test_subset_labels, mn_full_preds)\n",
      "cmp_eff['7 features 2500'] = metric(test_subset_labels, mn_small_preds)\n",
      "cmp_eff['7 features 5000'] = metric_5000(test_subset_labels, mn_small_preds)\n",
      "cmp_eff['no nmu 2500'] = metric(test_subset_labels, mn_no_nmu_preds)\n",
      "cmp_eff['no nmu 5000'] = metric_5000(test_subset_labels, mn_no_nmu_preds)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "2500HZ vs 5000HZ, no minpt cut"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "plot_eff(cmp_eff)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAA58AAAHyCAYAAABs5HrRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAAAAAAAAAAjLH/DzByV2GOdYlrAAAAAElFTkSuQmCC\n",
       "text": [
        "<matplotlib.figure.Figure at 0x33ef7690>"
       ]
      }
     ],
     "prompt_number": 46
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "metric_2500 = topo_utils.metric_generate(test.get_data(), 2500.0 / 1e6, test_empty_events, metric_plot=True)\n",
      "metric_3000 = topo_utils.metric_generate(test.get_data(), 3000.0 / 1e6, test_empty_events, metric_plot=True)\n",
      "metric_3500 = topo_utils.metric_generate(test.get_data(), 3500.0 / 1e6, test_empty_events, metric_plot=True)\n",
      "metric_4000 = topo_utils.metric_generate(test.get_data(), 4000.0 / 1e6, test_empty_events, metric_plot=True)\n",
      "metric_4500 = topo_utils.metric_generate(test.get_data(), 4500.0 / 1e6, test_empty_events, metric_plot=True)\n",
      "metric_5000 = topo_utils.metric_generate(test.get_data(), 5000.0 / 1e6, test_empty_events, metric_plot=True)\n"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "cmp2_eff = dict()\n",
      "cmp2_eff['2500'] = metric_2500(test_subset_labels, mn_full_preds)\n",
      "cmp2_eff['3000'] = metric_3000(test_subset_labels, mn_full_preds)\n",
      "cmp2_eff['3500'] = metric_3500(test_subset_labels, mn_full_preds)\n",
      "cmp2_eff['4000'] = metric_4000(test_subset_labels, mn_full_preds)\n",
      "cmp2_eff['4500'] = metric_4500(test_subset_labels, mn_full_preds)\n",
      "cmp2_eff['5000'] = metric_5000(test_subset_labels, mn_full_preds)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "All features and different HZ"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "plot_eff(cmp2_eff)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAA58AAAHyCAYAAABs5HrRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
       "text": [
        "<matplotlib.figure.Figure at 0x45364a10>"
       ]
      }
     ],
     "prompt_number": 47
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "import sklearn\n",
      "import sklearn.metrics\n",
      "quality_metric=sklearn.metrics.log_loss"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 53
    },
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "Checking every feature importance: by shuffling its values in test and and watching for changes in log loss metric"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "mn_no_nmu_importnace = mn_no_nmu.get_feature_importance_on_test(test, test_subset_labels, quality_metric)\n",
      "mn_full_importnace = mn_full.get_feature_importance_on_test(test, test_subset_labels, quality_metric)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "For mn_no_nmu"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "default_quality = quality_metric(test_subset_labels, mn_no_nmu_preds)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 64
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "for key, value in sorted(mn_no_nmu_importnace.iteritems(), key=lambda tup: tup[1], reverse=True):\n",
      "    print \"%s - %f\" % (key, value / default_quality - 1)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "fdchi2 - 0.259722\n",
        "sumipchi2 - 0.190386\n",
        "ipchi2 - 0.146206\n",
        "n - 0.040478\n",
        "ptau - 0.040222\n",
        "mcor - 0.032083\n",
        "sumpt - 0.023518\n",
        "fdr - 0.022169\n",
        "m - 0.010359\n",
        "vchi2 - 0.009979\n",
        "nhlt1 - 0.005061\n",
        "q - 0.000228\n",
        "nlt16 - -0.001107\n",
        "minpt - -0.009730\n",
        "eta - -0.011244\n"
       ]
      }
     ],
     "prompt_number": 70
    },
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "For mn_full:"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "full_default_quality = quality_metric(test_subset_labels, mn_full_preds)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 68
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "for key, value in sorted(mn_full_importnace.iteritems(), key=lambda tup: tup[1], reverse=True):\n",
      "    print \"%s - %f\" % (key, value / full_default_quality - 1)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "fdchi2 - 0.236595\n",
        "sumipchi2 - 0.184793\n",
        "ipchi2 - 0.128459\n",
        "ptau - 0.054585\n",
        "n - 0.038222\n",
        "mcor - 0.028923\n",
        "sumpt - 0.021191\n",
        "fdr - 0.017113\n",
        "vchi2 - 0.011007\n",
        "m - 0.008774\n",
        "nhlt1 - 0.003587\n",
        "nmu - 0.000096\n",
        "nlt16 - -0.000242\n",
        "q - -0.000634\n",
        "minpt - -0.010086\n",
        "eta - -0.012128\n"
       ]
      }
     ],
     "prompt_number": 71
    },
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "Bigger value - bigger increase in log loss - bigger feature importance for involved classifier."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [],
     "language": "python",
     "metadata": {},
     "outputs": []
    }
   ],
   "metadata": {}
  }
 ]
}